Industrial Application of Flotation Pulping Detection with Computer Vision - SME Annual Meeting 2026

Flotation performance is affected by upstream disturbances (e.g., mineralogy, water quality, flow rate) and controlled variables (e.g., pulp level and air addition). Process disruptions can lead to undesired conditions, such as pulp overflowing flotation cells (with insignificant/absent froth, and thus no valuable mineral concentration). Early detection and appropriate intervention of pulping events improves concentrator performance. An approach to real time flotation pulping event detection by means of convolutional neural networks is presented, including a case study of the industrial application of this technology on three flotation banks.